ZOTAC’s GeForce RTX 4090 4x 8-Pin-to-12VHPWR Cables Have a Limited Service Life: Up to 30 Connect/Disconnects

Tsing

The FPS Review
Staff member
Joined
May 6, 2019
Messages
11,075
Points
83
Enthusiasts who will be upgrading to one of ZOTAC's GeForce RTX 4090 graphics cards and plan to use the included 4x 8-pin-to-12VHPWR cable may want to plug and unplug their new hardware sparingly.

Go to post
 
I bet a lot of internal connectors are rated similarly to this and companies just don’t advertise it and you never hear about it or think to look for it
 
But short of benchmarking who is doing that much connecting/disconnecting?
I have no idea, but judging from the amount of broken HDMI connectors I see on repair channels, some people do like to connect/disconnect stuff a lot.
 
We're a long way from those horrible days of the 40/80 pin EIDE connectors but I've still had some reservations when upgrading cards in some of my rigs. Since I'm only recently beginning to incorporate liquid cooling, which I do believe helps in reducing overall temps in a case, I've been heavily focused on air solutions for everything up to now and noticed that my PCIe power cables are often almost completely inflexible after years of all that hot air being circulated around that has basically slow cooked them. It hasn't mattered which type of cable either and I've always felt a bit nervous when swapping cards with them. I'm all too aware of wire/metal memory so thinking about how much power and heat are involved with current cards is at the front of my mind when dealing with them.
 
I have no idea, but judging from the amount of broken HDMI connectors I see on repair channels, some people do like to connect/disconnect stuff a lot.
HDMI ugh! I love HDMI 2.1 but yeah, cables+connectors are an ongoing issue. I'm always saying that I could write a book about them with all the experiences I've personally had since the 1st gen.
 
I have no idea, but judging from the amount of broken HDMI connectors I see on repair channels, some people do like to connect/disconnect stuff a lot.
Yea people get utterly STUPID with how they treat cables and connectors outside of a case. I had to tell my wife.. "That is not meant to bend that way... STOP doing that you will ruin the port or the cable." Thankfully she took it to heart... now asks me when I'm doing wiring stuff if that's too much of a bend.. lol.

But the utter TWEAK that people do to cables is criminal. I want to make some sort of rigid spine like wrap for cables that makes audible cracking noises when people try to tweak it too far.
 
I've been heavily focused on air solutions for everything up to now and noticed that my PCIe power cables are often almost completely inflexible after years of all that hot air being circulated
Well, the power cables themselves get plenty hot - current flow generates heat because of internal resistance. Especially modern systems where we are looking at 250W+ CPUs and 450W+ GPUs -- that's why they have so many parallel connections (PCIe, EATX connectors, etc), to try to keep the current down in any single cable run.

It's probably that moreso than just the hot air; the air rarely gets ~that~ hot inside a case, but wire can get to over a 1,000F if you let it (it could technically get hotter but stuff starts melting past that)

The best examples are those poorly mated molex connectors that melt or catch on fire.
 
But short of benchmarking who is doing that much connecting/disconnecting?
It stacks up quickly. I was fiddling with my backup PC today, and I had to connect it, then disconnect it about 6 times before everything seemed to work as intended (well, almost). Sometimes everything works on the 1st try, but that's rare.
 
I doubt the thing has a counter on it and it will instantly fall apart at disconnect #31. But yeah, just the fact they published it makes me think it's more fragile than a typical connector
 
I doubt the thing has a counter on it and it will instantly fall apart at disconnect #31. But yeah, just the fact they published it makes me think it's more fragile than a typical connector
when you consider the amount of power that has to flow through those connections... that will degrade the metal, add on top of that the friction of uncommonly tight connections as tend to be used these days.... it makes sense.
 
That's surprising.

That said, I don't plan on relying on adapters. I've made that mistake before, and it was a fiery one.

If I go for one of these GPU's with their newfangled power plugs, I'll probably try to buy an official cable from my PSU manufacturer, to go straight from the modular PSU to the GPU instead.
 
We need a new standard, a new plug, more power, but did they have to design the worse plastic housing they could? I mean smaller pins, closer together, weaker plastics, and of course the darn things are melting.
 
There was no need for a new standard, just better engineering so the damned card doesn't pull 250W on one pin and 80W on the other.
New standard does nothing to fix that. All the new standard does is limit the power draw on startup and according to PSU specs. Which is completely redundant and stupid. Presumably the people buying 4090s have the PSU capacity to handle it.

The observed melting didn't occur because of using adaptors, but because of bad design, the risk is exactly the same if you use a cable directly from the PSU to the GPU.
 
when you consider the amount of power that has to flow through those connections... that will degrade the metal, add on top of that the friction of uncommonly tight connections as tend to be used these days.... it makes sense.
What would've made sense is having larger plugs so the contact area is larger. This is the apple way of designing things, dancing on the edge of the precipice. By using the least possible margins, like putting a high voltage LCD power pin next to a data pin going directly to the GPU die.
 
The more I read and watch videos on this the less I'm comfortable with ATX 3.0. Who thought it was a good idea to pull so much power through such tiny pins? And now I'm hearing that if you don't have an ATX 3.0 PSU with the 4 pin data connector that the GPU can inadvertently pull it's full power through 1 of the 3-4 leads on the adapter plug.

There are going to be fires. There are going to be lawsuits. Guaranteed.

I predict this power connection format will be short lived and updated sooner than we all think.
 
The more I read and watch videos on this the less I'm comfortable with ATX 3.0. Who thought it was a good idea to pull so much power through such tiny pins? And now I'm hearing that if you don't have an ATX 3.0 PSU with the 4 pin data connector that the GPU can inadvertently pull it's full power through 1 of the 3-4 leads on the adapter plug.

There are going to be fires. There are going to be lawsuits. Guaranteed.

I predict this power connection format will be short lived and updated sooner than we all think.
I was thinking this way myself. The power draw on these components is starting to get out of hand.
 
I'm not an authority on the subject of cables and connectors, but a durability rating of 30 mating cycles seems typical of the connector family to which the 6- and 8- pin PCIe power connectors belong, which for Molex I believe would be the Mini-Fit Jr. or Mini-Fit PCI Express series of connectors (unsure of the branding).

That's just what I found from poking around Molex's site (yeah, I know other brands exist).

(I opened way too many tabs, as usual.)

And the 30-max ratings I saw weren't even for a cable assembly as a whole, so treat it as an upper bound when connecting and disconnecting devices. The computer and all nearby devices will explode once any connector exceeds its limit. 💥 (They have internal counters!)

High durability options may exist. Someone with contacts in the power supply industry should ask about cable durability.

Addendum: I see an option for 100 and 250 mating cycles (tin and gold plated terminals, respectively), but didn't verify compatibility.
 
Become a Patron!
Back
Top